Research On High Concurrency Service Model Based On Nginx

The rapid development of the Internet,the growing number of users and the endless stream of Internet marketing activities have made the pressure on the server more and more.The adoption of the micro-service architecture puts higher requirements on the performance of the server,and the construction of the server cluster is more and more complicated.The load balancing technology has always been the focus of research.The mainstream load balancing technology focuses on solving the problem of task scheduling balancing,and does not pay much attention to the projects running on the server.From the perspective of the project,it is important to study the load balancing problem of service-oriented scheduling.This thesis mainly studies the scenario of network marketing activities under the "Internet+" environment by studying the commonly used cluster load balancing technology,and proposes a load balancing algorithm based on prediction threshold,and designs a high concurrent service scheduling model architecture.At the same time,the Nginx server was studied,and its architecture principle and strategy for coping with high concurrency were analyzed.This paper focuses on two types of scientific issues:?Cluster load check problem with high concurrency;? Service scheduling problems in clusters under complex conditions.The main research work of this paper is divided into the following aspects:1)Research and analysis of high concurrency scenarios in the "Internet+" environment.From the analysis of the process of Internet activities,the objective causes of high concurrency of Internet activities are analyzed.For the server cluster,the specific factors affecting the high concurrency ability are analyzed,and the countermeasures to improve the concurrency ability are pointed out.2)Scheduling and inspection model of high concurrent services under the network marketing scenario.Based on the theoretical model of game theory and the analysis of Internet marketing scenarios,a high-concurrency server cluster load verification model based on game theory is proposed.Based on this,the high concurrent service scheduling model architecture is further studied to solve the high concurrency situation.The service scheduling problem underneath provides a strong support.3)A dynamic weight load balancing algorithm based on prediction threshold is proposed.In-depth study of the current server load balancing algorithm,combined with the Internet marketing scenario,puts forward the shortcomings and advantages of the algorithm itself.Based on this,a dynamic weight load balancing algorithm based on predictive threshold is proposed.Dynamically evaluate server performance,calculate server weight based on evaluation results,and introduce thresholds for reaction cluster load balancing and status to avoid server jitter caused by frequent modification of weights;use Holt-Winters to predict traffic and predict results The threshold is used to implement load balancing in a complex load environment to achieve optimal load balancing.4)Implementation and verification based on microservice platform.The micro-service architecture is used to build a simulation experiment environment,and compared with multiple algorithms in the experimental environment,the algorithm and model are verified,and the experimental results are analyzed.Further deploying models and algorithms in the actual production environment,and simulating real user requests through Ali PTS,performing full link pressure measurement on the entire system to further verify the effectiveness of the algorithm in practical applications.The thesis conducts in-depth research on the high concurrency in the Internet marketing scenario.Based on the theoretical basis of game theory,this paper proposes a high concurrent service scheduling model,and studies the core algorithm,and proposes a dynamic weight load balancing algorithm based on prediction threshold..It can effectively cope with high concurrent Internet marketing scenarios,prompt cluster throughput,reduce average response time,and increase request success rate.The results of the thesis have been used in the actual production environment of an enterprise,effectively improving the quality of service and reducing the server cost of the enterprise.